1. Field of the Invention
The present invention relates to an image processing apparatus configured to perform image processing for a plurality of images, and a control method of the image processing apparatus.
2. Description of the Related Art
In recent years, an increase in resolution of a display (for example, 4096×2160 pixels) is promoted, and also a movie content with high resolution is created. There has been developed a technique (for example, see Japanese Patent Laid-Open No. 2009-260550) that, if an image processing capacity of a single image processing large scale integration (LSI) is not enough, divides an input image into a plurality of divided images and performs image processing for the plurality of divided images simultaneously by a plurality of image processing LSIs. Also, there has been developed a technique (for example, see Japanese Patent Laid-Open No. 2008-172410 and No. 2000-69293) that, when an input image is divided into a plurality of divided images and image processing is performed, makes connected lines in boundary portions of the plurality of divided images less noticeable.
FIGS. 10A and 10B are illustrations for explaining writing processing and reading processing by an image processing apparatus of related art. An input image is divided into two divided images D11 and D12, and the divided images D11 and D12 are written on a memory. During writing, the divided images D11 and D12 are written on the memory toward a lower side on a horizontal line basis. Since two image processing LSIs perform parallel processing, lines L1, L2, L3 . . . of the divided image D11 and lines L1, L2, L3 . . . of the divided image D12 can be written simultaneously.
During reading, an end portion d121 of the divided image D12 is added to the divided image D11, and the result is read as a read image D21. The read image D21 is read from the memory toward the lower side on a horizontal line basis. Also, an end portion d111 of the divided image D11 is added to the divided image D12, and the result is read as a read image D22. The read image D22 is read from the memory toward the lower side on a horizontal line basis. Since the two image processing LSIs perform parallel processing, lines L1, L2, L3 . . . of the read image D21 and lines L1, L2, L3 . . . of the read image D22 can be read simultaneously.
FIG. 11 is an illustration for explaining a writing timing and a reading timing by the image processing apparatus of the related art. The divided images (write images) D11 and D12 are written on the memory in synchronization with a vertical synchronizing signal Vsync1 for writing. In particular, the divided images D11 and D12 are written on the memory on a line basis in response to a synchronizing pulse at a time t11.
The read images D21 and D22 are read from the memory in synchronization with a vertical synchronizing signal Vsync2 for reading. In particular, the read images D21 and D22 are read from the memory on a line basis in response to a synchronizing pulse at a time t51 after writing of the divided images D11 and D12 is completed.
The read images D21 and D22 read from the memory are processed with parallel image processing by the two image processing LSIs, then the added portions d111 and d121 are cut, and the results are connected together. As described above, by adding the end portions d111 and d121, which overlap each other, to the divided images D11 and D12, and reading the results, an output image, which has a less noticeable connected line in a boundary portion between the divided images D11 and D12 and has less deterioration in image quality, can be obtained.
However, in the related art described in FIGS. 10A, 10B, and 11, since the end portion d111 of the divided image D11 is added to the read image D22, the read image D22 cannot be read from the memory only after the entire divided image D11 is written on the memory. Hence, the reading timing of the read images D21 and D22 is delayed relative to the writing timing of the divided images D11 and D12 by about a single field. Owing to this, when image processing is performed for a plurality of divided images, a delay time of image output is increased due to delay of reading.